基于BP神经网络的河道水位推算模型研究

在分析影响河道水位因素的基础上，采用基于梯度下降算法的BP神经网络模型推算河道水位，同时采用传统的上下游水位线性相关方法进行水位推算。结果表明，在具有较长期实测资料的情况下，BP神经网络模型具有很高的精度，若要考虑更多相互独立影响因素的非线性作用，应相应增加输入样本数。文章最后提出了进一步提高模型精度的几点设想。

Calculation Model for River Water Level Based on BP Neural Network

On the basis of the analysis of factors affecting the river water level, the BP neural network model,based on the gradient descending algorithm,is used to calculate the river water level.A comparison with the results of the linear correlation method proves that the BP model is of high precision when long term observed data are available.If more independent affecting factors are to be considered for their nonlinear effect,more input water level series are needed.The BP model is a better way than the linear correlation method to extend the water level series.Finally,some suggestions for further improvement of model precision are proposed.